ERS Precise Orbit Determination

Frequently Asked Questions (FAQ)


This page is intended to help you in case you have a question about our orbits. This page contains many of the Frequently Asked Questions. I have tried to organise the questions by topic, but maybe it is best to use the Find tool of your browser to look for some keywords.
This page is under construction

Last update: Wed Feb 11 14:37:44 CET 2004


Content

Data availability
Q: Are there orbits for ERS-1 during October 1997?
Q: Why are there no precise ERS-1 orbits since June 1996?

Orbit quality
Q: As far as I could learn from your homepage, the accuracy of the DEOS precise orbits is about 5-6 cm radially for ERS-1. Can you tell me the accuracy for ERS-2 (until Jan. 2000)?
Q: How much accuracy degradation can be expected for the fast-delivery and the preliminary orbits compared to the precise orbits?
Q: Since you do not provide precise orbits for ERS-1 in 1997, can you estimate the error of the CEOS-orbits?
Q: I need the "best" orbit information because I do some interferometric analyses with these ERS data and the accuracy of the baseline (vector between satellite acquisition positions) is a very critical point. The difference between the DEOS precise orbits and the information in the CEOS headers is several metres. Which orbit can I trust?
Q: What happens in getorb when there is a manoeuvre?
Q: The getorb routine sometimes says "outside precise range". What does this mean?
Q: How good or bad are the orbits "outside precise range"? Would they have slope errors of greater than 2 microradian?

Phases, cycles, and passes
Q: How are the passes, cycles and phases defined?
Q: Are the repeat cycles of ERS-1 all exact multiples of one day or 3.xx, 35.xx, 168.xx days?
Q: Do you have a table showing the start of each cycle?

Position and velocity
Q: Is there a simple way to retrieve XYZ velocities from the software?

Software installation
Q: Getorb does not work under Cygwin or Mac OS X (using fink).

Data availability

Q: Are there orbits for ERS-1 during October 1997?
No. The reason for this is that the satellite was not tracked by any satellite laser systems between July 1996 (when ERS-1 was put into hybernation) and July 1998, when it was officially revived. During some peiods, including October 1997, ERS-1 was revived, but tracking of the satellite by laser systems (a necessity for precise orbit determination) was never requested for that period. Therefore, it will not be possible for us, or anybody, to provide precise orbits for ERS-1 for the period between July 1996 and July 1998.

Q: Why are there no precise ERS-1 orbits since June 1996?
The reason why there are no more precise ERS-1 orbits is the fact that the radar altimeter is not working. We need this instrument to compute orbits for ERS-1 with a good precision. In particular for the period of October 1997, we will not be able to compute any orbits for ERS-1 since the satellite was also not tracked by any laser tracking station. Laser tracking recommenced in July 1998.

However, we have started since a while to generate operational orbits again for ERS-1, especially since the large interest we received from them after the unfortunate earthquakes in Turkey. Sadly, we can not expect these orbits to be as precise as they were before 1996. To retrieve the operational orbits please follow the links to "fast-delivery orbits for ERS-1".


Orbit quality

Q: As far as I could learn from your homepage, the accuracy of the DEOS precise orbits is about 5-6 cm radially for ERS-1. Can you tell me the accuracy for ERS-2 (until Jan. 2000)?
ERS-2 orbit accuracies are the same as for ERS-1: about 5 cm radial and about 15 cm in along- and cross-track direction. After Nov 1999 the quality degrades from time to time in the event of solar flares (periods of high solar activity).

We are working on improvements of the drag modelling, which will be particularly benificiary for those periods.

Q: How much accuracy degradation can be expected for the fast-delivery and the preliminary orbits compared to the precise orbits?
Any answer to this is, of course, a general one. There will always be cases when the reality is different from general practise. So in general, I can say that our fast-delivery orbits are remarkably close to our final orbits. The differences between the fast-delivery and precise orbits have even reduced during the ERS-2 mission, and are now roughly 2 cm RMS in radial, 8 cm in along-track and 6 cm in cross-track direction. Obviously, there are exceptions, especially near orbital manoeuvres, when the lack of tracking data and time constraints on the orbit production lead to larger errors.

More care has been spent on the preliminary orbits. They differ by approximately 1, 4 and 3 cm RMS in radial, along-track and cross-track direction.

Q: Since you do not provide precise orbits for ERS-1 in 1997, can you estimate the error of the CEOS-orbits?
Yes, orbit determination for ERS-1 in 1997 is imposible. I'm not sure how well the CEOS orbits do, but they will have a similar quality as earlier periods. Generally, the errors in those orbits are between 2 and 10 metres.

Q: I need the "best" orbit information because I do some interferometric analyses with these ERS data and the accuracy of the baseline (vector between satellite acquisition positions) is a very critical point. The difference between the DEOS precise orbits and the information in the CEOS headers is several metres. Which orbit can I trust?
The orbit information stored in the headers is a very rough guess of where the satellite is. As far as I have been told it is based on the propagation of the state vector (position and velocity) at the latest ascending equator crossing. And even that state vector is based on a very crude orbit determination. Needless to say the orbit in the header can be wrong by several meters.

The DEOS orbits are accurate to about 4-5 cm in vertical and 10-15 cm in horizontal directions (both cross-track and along-track). Some of this error is, however, geographically correlated. This means that it is repeatible from cycle to cycle (and will be the same for ERS-1 and ERS-2 along the same gound-track). So this part (which can de several centimeters) cancels in the orbit differences that you compute for InSAR applications.

Q: What happens in getorb when there is a manoeuvre?
All ODRs are continuous in the sense that there are no gaps. When there is an in-plane manoeuvre, we calculate right through them. This ensures also a continuous orbit in the sense that there are no magic jumps. However, when there is an out-of-plane manoeuvre (every 2-3 months) we can not calculate through it. That is not so much because we can't handle accelerations in out-of-plane directions, it is simply that the delta-V's are so large and uncertain that the prerequisit for linear least squares estimation (small corrections) fails. Thus we cut the start and stop the orbit computation at out-of-plane manoeuvres and then paste the two pieces together to form one ODR. There will be a discontinuity at that point, but the interpolation routine is able to go through that, although it will produce worse orbits around that point.

Secondly, there is generally no tracking data between the out-of-plane manoeuve and the next in-plane manoeuvres (a few hour later). In between the orbit is seriously degraded, but since the altimeter is often switched of during that period, this is not always a problem.

Look at the manoeuvre files for a history of ERS-1 and ERS-2 manoeuvres.

Q: The getorb routine sometimes says "outside precise range". What does this mean?
the file you want to use for the interpolation. If you specify in the getorb call the directory name instead, it should automatically jump to the appropriate ODR file when it gets "outside precise range".

This, by the way, only means that you request the interpolation of the orbit outside the middle 50 or 51 revolutions (roughly 3.5 days) of the ODR file. The ODR file contains 5.5 days of orbits, but only the middle part should be used, since the outer limits get less accurate. It's the free end of the solution, like a plank suspended over a couple of trusses. The ends that stick out are much more flexible than the inner part.

Q: How good or bad are the orbits "outside precise range"? Would they have slope errors of greater than 2 microradian?
If you only go slightly beyond the middle 3.5 days, you're fine. Its basically the outer 6 hours or so that get slightly less accurate, but not even close to the 2 microradian level (even a 1 metre orbit error is only 0.025 microradian).

Phases, cycles, and passes

Q: How are the passes, cycles and phases defined?
A list of definitions and the start times of all the passes and cycles can be found on the web page ERS-1 and ERS-2 Cycle and Orbit Start Times.

Q: Are the repeat cycles of ERS-1 all exact multiples of one day or 3.xx, 35.xx, 168.xx days?
Yes, nominally the repeat cycles are all exact multiples of 1 day. The cycles of Phases A, B and D (Commissioning Phase and the First and Second Ice Phase) are exactly 3 days long. During the two Multi-disciplinary phases C and G the cycles were exactly 35 days. The two Geodetic Phases contained only about 1 cycle each with a length of exactly 168 days. In reality the length of the cycles can slightly differ from the nominal one due to orbit decay. However this difference is never more than a few seconds.

Click on for more info on ERS repeat cycles and phases and a full lists of start times of each cycle.

Q: Do you have a table showing the start of each cycle?
Yes, this information is all available on the web page ERS-1 and ERS-2 Cycle and Orbit Start Times.

Position and velocity

Q: Is there a simple way to retrieve XYZ velocities from the software?
In the main program (getorb_main.f) you can see how to compute XYZ positions (XYZpos). Since the orbit is very nearly a straight line during one second you can approximate to a high degree of precision the XYZ velocity (XYZvel) as follows:

XYZvel (t) = XYZpos (t + 0.5sec) - XYZpos (t - 0.5sec)


Software installation

Q: Getorb does not work under Cygwin or Mac OS X (using fink).
Reportedly, lseek has a problem on both these platforms. It is easy to solve by adding the line
#include <unistd.h>
near the head of fastio.c.

Status | Orbits | Details | FAQ | Literature | Results |
ERS orbits | Links | DEOS Home

DEOS Home This page was created by Remko Scharroo, remko.scharroo@noaa.gov
It is currently maintained by Eelco Doornbos, e.n.doornbos@tudelft.nl